Video footage grows exponentially given the advancement in sensor technology. Law enforcement agencies across the nation are now more and more equipped with body cameras—a tremendous data source addition to the existing in-vehicle dash cameras. Due to the Freedom of Information Act requests continue to climb as public demands more and more transparency from law enforcement. There is an imminent need for the authority to rapidly redact video footage before releasing them to the public for privacy protection.
There are many redaction solutions in the market for static media such as images and documents. Image/document redaction techniques are not applicable due to the large volume and the highly dynamic content of videos. Based on the same argument, manual redaction is just not feasible for videos—e.g., a 5-minute video with 30 fps has 9000 frames. Recently due to the rapid spread of body worn camera usage, it has been proposed to use visual tracking technologies for redaction.